Method for diagnosing type of leakage of drilled well based on neural network fusion technique

Abstract

The invention discloses a method for diagnosing the type of leakage of a drilled well based on the neural network fusion technique. The method for diagnosing the type of leakage of the drilled well based on the neural network fusion technique comprises the following steps that S1, a fusion neural network structure is determined, and a well leakage type diagnosis parameter space is established; S2, data are preprocessed; S3, data which are normalized are input into all neural networks; S4, data infusion is conducted on output values of all the neural networks according to multiple neural network fusion algorithms and then a well leakage type diagnosis result is obtained finally. The step S1 comprises the sub-steps that S11, the fusion neural network structure is selected; S12, network training is conducted; S13, the networks are checked, and the well leakage type diagnosis parameter space is established. According to the method for diagnosing the type of leakage of the drilled well based on the neural network fusion, the multiple neural networks are used for processing data, the multiple neural network output results are processed through the data fusion algorithm, the performance is high, and the result is highly reliable.

Description

technical field [0001] The invention relates to a method for diagnosing the type of lost circulation in drilling based on neural network fusion technology. Background technique [0002] Drilling engineering is the primary technology for oil and gas exploration and development. During drilling operations, downhole complex situations and drilling accidents have been threatening the entire drilling process, which not only have a serious impact on well construction quality and drilling speed, but also greatly reduce exploration benefits. With the rapid development of drilling technology, the current goal of the petroleum industry is to reduce the incidence of drilling accidents and drilling costs, and improve drilling quality and efficiency. [0003] At present, during drilling operations, accident diagnosis and identification of abnormal conditions mainly rely on the naked eye observation of on-site operators and engineering technicians, combined with past experience and drill...

AI Technical Summary

Problems solved by technology

[0024] In short, there are many researches on the diagnostic models of pipe sticking and blowout accidents at home and abroad, but there are relatively few researches on the diagnostic models of lost circulation types, and the existing lost circulation studies still have the following deficiencies: (1) Basic parameters The inaccuracy is mainly due to poor measurement technology, limited experimental means, and imperfect theoretical calculation models; (2) Theoretical research on lost circulation is not perfect; (3) The mechanism of lost circulation is still in the qualitative or semi-quantitative description stage, lacking More accurate quantitative evaluation

Images